linux-kernel - Re: [PATCH] arm64: kernel: memory corruptions due non-disabled PAN

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <CA+CK2bDvp=RR7A5J097W3tr2QG_cOZgDMno388EXGAYW6+Q1Mg@mail.gmail.com>
Date:   Wed, 20 Nov 2019 14:52:06 -0500
From:   Pavel Tatashin <pasha.tatashin@...een.com>
To:     Mark Rutland <mark.rutland@....com>
Cc:     James Morris <jmorris@...ei.org>, Sasha Levin <sashal@...nel.org>,
        LKML <linux-kernel@...r.kernel.org>,
        Catalin Marinas <catalin.marinas@....com>,
        Will Deacon <will@...nel.org>, steve.capper@....com,
        Linux ARM <linux-arm-kernel@...ts.infradead.org>,
        Marc Zyngier <marc.zyngier@....com>,
        James Morse <james.morse@....com>,
        Vladimir Murzin <vladimir.murzin@....com>,
        Thomas Gleixner <tglx@...utronix.de>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        allison@...utok.net, info@...ux.net, alexios.zavras@...el.com
Subject: Re: [PATCH] arm64: kernel: memory corruptions due non-disabled PAN

On Wed, Nov 20, 2019 at 2:46 PM Pavel Tatashin
<pasha.tatashin@...een.com> wrote:
>
> > > I see that with CONFIG_ARM64_SW_TTBR0_PAN=y, this means that we may
> > > leave the stale TTBR0 value installed across a context-switch (and have
> > > reproduced that locally), but I'm having some difficulty reproducing the
> > > corruption that you see.
> >
> > I will send the full test shortly. Note, I was never able to reproduce
> > it in QEMU, only on real hardware. Also, for some unknown reason after
> > kexec I could not reproduce it only during first boot, so it is
> > somewhat fragile, but I am sure it can be reproduced in other cases as
> > well, it is just my reproducer is not tunes for that.
> >
>
> Attached is the test program that I used to reproduce memory corruption.
> Test on board with Broadcom's Stingray SoC.

I forgot to remove from repro.c some of the stuff that I used for debugging:
get_pa() and sched_setaffinity() stuff are not needed.

>
> Without fix:
> # time /tmp/repro
> Corruption: pid 1474 map[0] 1488 cpu 3
> Terminated
>
> real    0m0.088s
> user    0m0.004s
> sys     0m0.071s
>
> With the fix:
>
> # time /tmp/repro
> Test passed, all good
> Terminated
>
> real    1m1.286s
> user    0m0.004s
> sys     0m0.970s
>
>
>
> Pasha